• MTO Home
  • Current Issue
  • Previous Issues
    By Author By Volume       
  • Submit
    Submission Guidelines Submit Online
  • Jobs
    Current Job Listings Submit Job Listing
  • Dissertations
    All Dissertations New Dissertations List Your Dissertation
  • Events
  • Webinars
  • About
  • Links
  • SMT

Volume 18, Number 3, September 2012
Copyright © 2012 Society for Music Theory

“Possible Paths”: Schemata of Phrasing and Melody in Charlie Parker’s Blues

Stefan C. Love


KEYWORDS: Charlie Parker, blues, improvisation, schema, formula, phrasing

ABSTRACT: This paper proposes a schematic approach to analyzing Charlie Parker’s improvisations on the twelve-bar blues. Schemata are pre-learned, recurring solutions to the problems of high-speed improvisation. Phrasing schemata solve the meter problem: they are templates for the chorus-level organization of phrases. Parker employs five phrasing schemata, each of which organizes the blues’ twelve measures differently. Melodic schemata solve the harmony problem: they are stepwise paths that Parker follows in the two different “Zones” of the blues’ harmonic structure. Drawing on a small repertory of versatile schemata in both domains, Parker can compose intricate, varied melodies in the act of performance. After presenting the schemata, the paper concludes with a schematic analysis of a three-chorus solo.

Received July 2012

PDF version
 

“I now understand improvisation as the real-time yet pre-heard—and even practiced—choice among possible paths that elaborate a pre-existing structure, using familiar patterns and their familiar combinations and embellishments.”—Steve Larson (2005, 272)

Introduction

[1.1] Charlie Parker was a master improviser. His extraordinary oeuvre carries an implicit question: how did he do it? How did he create such well-crafted melodies in the very act of performance? This is “the problem of improvisation”—a problem for both the improviser and the analyst.(1) Inspired by the epigraph from Steve Larson, this paper posits one kind of solution for Parker’s improvisations on the twelve-bar blues. I suggest that Parker developed two different sets of “possible paths” through the blues: phrase structures, to solve the problem of how to place phrases against a fixed meter; and melodic paths, to solve the problem of creating coherent melodies against a fixed harmonic structure. I call these paths schemata of phrasing and melody.

[1.2] The term “schemata” is most familiar from the work of Robert Gjerdingen on the galant style (2007). The galant composer confronted a similar “problem” to the bebop improviser: high-speed composition. Gjerdingen’s schemata are recurring patterns of scale-degrees. Through artful elaboration and combination of pre-learned schemata, galant musicians could compose efficiently and expressively. Without using this term, jazz scholars have similarly speculated that the improviser draws on a pre-learned repertory of (pitch-based) musical materials. An improviser imprints this repertory through practice, so that it spontaneously emerges in a solo. My schemata are based on this same view of improvisation. They are Parker’s raw improvisational materials, highly flexible in their details and application, and they facilitate efficient composition (to put it mildly). They inhabit two domains: melody, in the form of recurring sequences of scale-degrees (very similar to Gjerdingen’s); and phrasing, in the form of recurring relationships between phrasing and meter. The interaction of these domains further increases Parker’s economy of means: the same melodic schema sounds quite different when laid over two different phrasing schemata.

Example 1. The metrical-harmonic structure of a typical bebop blues (for simplicity, sevenths are omitted)

spacer

(click to enlarge)

[1.3] The twelve-bar blues has been a basis for jazz improvisation perhaps longer than any other form. Example 1 gives a metrical-harmonic outline of a bebop blues.(2) There are three four-bar hypermeasures, describing a single harmonic phrase. The first hypermeasure prolongs tonic, the second hypermeasure moves to the subdominant and then back to tonic, and the third hypermeasure presents a ii–V–I cadence.(3) Harmonic substitution and interpolation are common; but alterations preserve the opening establishment of tonic, measure 5 motion to the subdominant, and cadence terminating in measure 11.(4) Charlie Parker had a particularly intimate relationship with the blues: it represents around a quarter of Parker’s recorded output, and it likely constituted an equivalent portion of his practice time (Martin 1996, 3). It is only natural that he would develop shortcuts to improvising on this familiar form.

[1.4] To identify schemata, I sought recurring patterns in a sample of Parker’s blues. This sample included thirty-nine recorded performances totaling 156 improvised choruses, dating from between 1944 and 1953, after Parker’s style had reached maturity (Owens 1974, vol. 1, 5). These thirty-nine performances fit two criteria, one practical, the other theoretical: a recording and published transcription were readily available;(5) and the tempo was at least 160 quarter-notes per minute.(6) I speculate that the patterns in this sample would be found throughout Parker’s blues performances.

The Solo as Interpretation(7)

[2.1] The form of a bebop performance is often described as following a theme-and-variations model; this familiar analogy needs refining. The technique of “bebop variation” might be better understood if we distinguish how the theme’s three elements of meter, harmony, and melody are treated during the solos. The metrical structure of a solo, from the level of the quarter-note beat through the level of the chorus, is entirely based on that of the theme. During the performance, the musicians and experienced listeners maintain continual awareness of this structure, anticipating the downbeats of each measure, hypermeasure, and chorus. Any metrical dissonance is superimposed on this structure without threatening it. (Indeed, true disruption of the meter, in the form of added or subtracted beats or measures, is viewed as a mistake.) The ensemble or soloist can alter the harmonic structure, but they tend to preserve the harmonic middleground. In contrast with both of these elements—rigid meter, elastic harmony—the soloist’s melody is free.

Example 2. The freedom of the improvised melody

spacer

(click to enlarge and listen)

[2.2] How is the soloist’s melody free? Though experts differ on this point, it is safe to claim that Parker usually avoids obvious thematic reference, and his practice is not unusual.(8) Furthermore, the phrase structure of the melody need not follow the theme or the hypermeter. Example 2 illustrates this: through rests, Parker creates the phrase structure 2/8/2, with the last phrase spilling into the next chorus. I call such phrase structures “dissonant,” because the phrasing contradicts the theme’s 4/4/4 hypermeter (and the 4/4/4 phrase structure of the theme’s melody). (By extension, phrase structures in alignment with the meter are “consonant.”) Example 2 also demonstrates the soloist’s freedom to deviate even from the theme’s essential harmonic structure (though this is rarer): Parker follows the theme’s pre-dominant harmony in measure 9, but ignores the rest of the cadence. Instead, the final two measures of the chorus are filled with highly altered dominant harmony, as a long lead-in to the next chorus. Parker suppresses the theme’s final tonic resolution altogether.

Example 3. Different cadences over the same ostinato

spacer

(click to enlarge and listen)

[2.3] Because of the melody’s freedom, the solo is best understood not as a variation on the theme, but as an interpretation of it: a solo may follow the theme’s harmony, melody, or phrase structure, or it may not. To the extent a solo deviates from its theme, it presents a new, revised structure. As another example, soloists commonly transform a theme’s half-cadence and restart—“→V || I→”—into a full cadence by playing across the dominant and ending a phrase on the tonic: “→V→I||”. Such flexibility represents a major difference between bebop and common-practice strophic variation. Whereas the bebop soloist may reinforce, ignore, or revise the theme, in strophic variation, melody, harmony, and meter work in concert to elaborate the theme or an underlying common structure. The resemblance is much stronger between bebop and continuous (ostinato) variation. Consider the excerpt from “Dido’s Lament” shown in Example 3: in measures 32, 39, and 44, against an unwavering ostinato, the voice variously establishes a half-cadence, an elided cadence, and a full cadence. The ostinato is a foil to the vocalist, who may reinforce or revise its “natural” grouping and harmonic structure; the aggregate effect varies from cycle to cycle. Parker’s relationship to the theme is identical. The interpretive view of improvisation leads me to abandon the assumption of a consistent chorus-level structure. Instead, schemata describe the structures Parker actively creates.

Phrasing Schemata(9)

[3.1] Unlike thirty-two bar themes with eight-bar sections, the blues has no metrical or formal level between the four-bar hypermeasure and the complete chorus. The blues’ brevity increases the risk that a multi-chorus solo will become tedious. Furthermore, the blues’ three hypermeasures oppose the duple construction typical of eight- and sixteen-bar sections. These features heighten the improviser’s “meter problem”: how to organize phrases in a varied way over a repetitive metrical structure, without the need for continual invention.

[3.2] I propose that Parker solved the problem with phrasing schemata. Each is a “path” through the metrical structure of one blues chorus, an adaptable template for where phrases begin and end. For example, one common schema is 8/4: an eight-measure phrase followed by a four-measure phrase (the phrases may be further divided). By “phrase,” I do not mean a tonally defined phrase terminating in a cadence. Given a repetitive harmonic structure, the soloist has only limited control over those. Rather, I use the term in a deliberately colloquial sense: a phrase is a continuous musical gesture, surrounded by rests. These phrases are highly variable across choruses and performances. Soloists exert total control over their placement.

[3.3] Several authors mention phrasing as a broadly interesting aspect of jazz, though few explore it in detailed analysis.(10) One exception to this is Steve Larson. In his analyses, he comments on the metrical lengths of phrases and describes the effect of “sentences” and “reverse sentences” (1993, 292–93; 1996, 154; 1999a, 298–99). He implicitly follows the colloquial definition for phrase given above. Contemplating an extended metrical dissonance in a Bill Evans solo, Larson even points to the concept of the metrical schema: “To make such a journey through the ‘metric space’ of this piece . . . Evans must have known that metric space intimately, must have internalized its possible basic rhythmic paths securely, and must have developed many ways of traveling those paths flexibly and fluently” (2005, 255). Though phrasing schemata are a more general kind of path, the core idea of “rhythmic paths” is common to both approaches.

Example 4. The phrase hierarchy and the “primary division”

spacer

(click to enlarge and listen)

[3.4] To refine the phrase’s colloquial definition, first I invoke three musical features that suggest a phrase division: inter-onset interval (IOI, or the time between subsequent attacks); melodic discontinuity; and proximity of a strong beat.(11) Taken together, these features establish the lowest-level phrase divisions in a chorus. Example 4 illustrates my approach. At the lowest level, we might identify six segments, separated in the example with dotted lines.(12) The three features of IOI, discontinuity, and strong beat, in combination with other factors, especially melodic parallelism, further imply a hierarchy of phrases above this lowest level. In Example 4, the segments in measures 1–8 are united by their approximate parallelism and the closure suggested in measure 7 by the return of spacer spacer over tonic; melodic continuity unites the segments in measures 9–11 (up-down contour and Db–C connection); and these groups of segments are separated by the chorus’s longest IOI in measures 7–9. The lowest level might be modeled 2/2/2/2/2/2; the next level up is 8/4 (numbers correspond approximately with length in measures). Like many choruses, this chorus has a single deepest phrase division, shown with a double bracket, which I call the “primary division.”

Example 5. The five phrasing schemata

spacer

(click to enlarge)

[3.5] Phrasing schemata name the highest level of phrase structure within the chorus. In the case of Example 4, the schema is 8/4. This designation glosses over the chorus’s lower-level subtleties but permits comparison with other choruses that fit the same pattern. In total, I identify five phrasing schemata: 4/4/4, 8/4, 4/8, 6/6, and Through-Composed (TC). Example 5 depicts the group. The 8/4, 4/8, and 6/6 schemata have a primary division: they create a level of phrase structure between the 4/4/4 hypermeter and the twelve-measure chorus. 4/4/4 and TC do not exhibit any obvious phrase structure beyond the hypermetrical level. The schemata are defined exhaustivelyin theory, every chorus can fit into one category or another, and most are easy to classify. However, when phrasing factors conflict, ambiguity can arise. Below, I discuss each schema in more depth and analyze two ambiguous choruses. Then I present data on schema frequency and syntax.

Example 6a. A typical example of the phrasing schema 4/4/4

spacer

(click to enlarge and listen)

Example 6b. A more complex example of 4/4/4, due to phrase subdivision, rhyme, and voice leading

spacer

(click to enlarge and listen)

Example 7a. A typical example of 8/4, with phrases undivided

spacer

(click to enlarge and listen)

Example 7b. A more dissonant example of 8/4, due to off-tonic ending in measure 8

spacer

(click to enlarge and listen)

Example 8a. A typical example of 4/8

spacer

(click to enlarge and listen)

Example 8b. A 4/8 chorus with a clear return to tonic in measure 7, heard as a passing chord

spacer

(click to enlarge and listen)

Example 8c. A 4/8 chorus with a late, extended tonic in measures 7–8, dividing the pre-dominant area (measures 5–9)

spacer

(click to enlarge and listen)

Example 9a. A typical example of 6/6

spacer

(click to enlarge and listen)

Example 9b. An unusual example of 6/6, due to long IOI in measures 8–9

spacer

(click to enlarge and listen)

Example 10a. A typical example of a Through-Composed (TC) chorus, due to phrase divisions in the “wrong” places

spacer

(click to enlarge and listen)

Example 10b. A TC chorus with no divisions at all

spacer

(click to enlarge and listen)

Example 11a. An ambiguous chorus: possible primary divisions are blurred by other factors

spacer

(click to enlarge and listen)

Example 11b. Another ambiguous chorus

spacer

(click to enlarge and listen)

[3.6] In the 4/4/4 schema, there are two phrase divisions of comparable significance at or near the downbeats of measures 5 and 9. The phrase structure follows the hypermetrical structure, and can seem square or predictable: overly consonant. Example 6a is typical: the metrical rhyme between the three phrases highlights their equal status. Metrical rhyme (“rhyme” hereafter) is a form of parallelism among phrases arising when they begin or end at (or near) the same relative point in a measure or hypermeasure (Love 2012, 25). It would not make sense to group only two of the three rhyming phrases at a higher level. Example 6b also suggests 4/4/4, though it exhibits more complexity. The first two phrases are divided, but not the third. The second and third phrases begin at the same point in their hypermeasures (“beginning-rhyme”), and the tense spacer spacer in measure 8 blurs these phrases into a single gesture of tension-release. However, none of these factors decisively groups an adjacent pair of four-measure phrases together, as would suggest 8/4 or 4/8; 4/4/4 seems the best analysis.

[3.7] We have already seen one instance of the 8/4 schema in Example 4; two more appear as Examples 7a and 7b (below, Examples 21, 23, and 25 also follow 8/4). In this schema, the primary division falls in measure 7 or 8, roughly aligned with the division between the second and third hypermeasures. The effect is consonant: a seven-measure elaboration of tonic, separate from the more energetic cadence. In Example 4 above, the constituent phrases are further divided into two-measure segments; in Example 7a, the phrases are undivided, unfolding as continuous gestures and separated by a long rest. This pattern is common at fast tempos. In Example 7b, the eight-measure phrase is divided, but the stepwise connection from F to Espacer across measures 4–5 links its two subphrases together. The faster rhythm of the final four-measure phrase sets it apart. However, compared to Examples 4 and 7a, the eight-measure phrase is open-ended: Parker extends it past the tonic in measure 7, to V/ii. (Cf. Example 6b, measure 8, from the same performance, where other factors establish 4/4/4.)

[3.8] The 4/8 schema is the counterpart to 8/4, grouping the chorus’s second two hypermeasures together. The primary division falls before the arrival of IV in measure 5, roughly in line with the first hypermetrical division. Example 8a is typical. Measures 5–11 unfold as a single phrase, with a subphrase division in measure 9. (The Aspacer in measure 12 is a pickup to the next chorus.) Note that Parker downplays the return to tonic in measure 7, so that the opening tonic prolongation seems only to occupy measures 1–4, with pre-dominant extending from measures 5–9.(13) The 4/8 schema places the resulting “extended pre-dominant” (measure 5–9) in a single phrase with the final cadence. The listener might perceive an early increase in tension, as the drive to the cadence begins four measures earlier than in the 8/4 schema. Example 8b is another realization of 4/8. Though the long IOI (measures 3–4) firmly establishes the primary division, Parker presents a clear return to tonic in measure 7 (anticipated in measure 6), unlike in Example 8a. This can be interpreted as a passing chord (along with V/ii in measure 8) within a larger motion from IV to ii. The melody’s propulsive energy makes it unlikely that measure 7 would be heard as a stable return to the opening harmony. Example 8c deviates further from the prototypical form of 4/8: a long descent from spacer to spacer in measures 7–8 (a delayed variant of the “Descent to spacer ” schema, discussed below) emphasizes tonic and suppresses V/ii. Although the long phrase still spans the pre-dominant area, such a pronounced return to tonic is unusual. As a result, one might hear two distinct, overlapping phrases: IV–I | I–ii–V (measures 5–8 | 8–10).(14)

[3.9] In the 6/6 schema, the primary division falls between the IV chord in measure 5 and the return to tonic in measure 7. This causes two kinds of dissonance: between six-measure phrases and four-measure hypermeasures, and between phrasing and middleground harmonic structure. Whether one hears an extended tonic in measures 1–7 or an extended pre-dominant in measures 5–9, the primary division divides this area. Parker tends to present 6/6 with little ambiguity, using a long IOI at the primary division. This is evident in Examples 9a and 9b. (Other 6/6 choruses are Examples 15, 22, and 24, below.) Example 9b includes a sizable IOI in measures 8–9, a rival to the primary division, and an unusual feature for a 6/6 chorus. Parker’s subdivision of each six-measure phrase into 2/4 reinforces the chorus-level 6/6 structure: the subphrase in measures 1–2 has its counterpart in measures 7–8, and the subphrases in measures 3–5 and 9–11 nearly rhyme (compare their starting and ending beats). But the two-bar displacement with the hypermeter makes this parallelism hard to hear.

[3.10] TC choruses do not fit into any of the other categories. They may have phrase divisions in the “wrong” places, as in Example 10a, where Parker divides the first and last hypermeasures (cf. Example 2 above: another 2/8/2 structure). The long phrase from measures 2–9 is balanced by a brief final phrase that restores order: note the end-rhyme between measures 9 and 11, creating a sense of closure. (This has a remarkable counterpart in the beginning-rhyme between measures 0 and 2!) At faster tempos, TC choruses sometimes have no phrase divisions at all, as in Example 10b. TC and 6/6 choruses generally sound more dissonant than 8/4 and 4/8, whose primary divisions align with a hypermetrical division.

[3.11] Example 11a lends itself to multiple interpretations. Its phrase divisions in measures 4 and 6 point to 4/8 or 6/6. However, Parker blurs both of these, such that neither is likely to suggest the depth of a true primary division. The rhyme between the endings in measures 2, 4, and 6 cuts across the division in measure 4: all three subphrases end on beat 6.5 of a two-bar hypermeasure. The beginning-rhyme between measures 1, 3, 5, and 7—all four phrases begin just before a two-bar downbeat—cuts across both divisions, as does the unified up-down contour of measures 3–9—note especially the link between the high F in measure 5 and E in measure 6. It would be hard to hear a primary division anywhere in this chorus. Ultimately, I would classify this as TC by default, though the “square” placement of its divisions distinguishes it from most TC choruses.

[3.12] In Example 11b, Parker blurs the phrase divisions more subtly. Again, 4/8 and 6/6 are plausible interpretations. However, the first phrase ends on a note of tension at measure 4: Bspacer , the flat seventh, and a relative high point. The transparent voice leading from this Bspacer to the A in measure 5 weakens the first division. The long IOI in measures 6–7 would initially indicate 6/6; but the parallelism between the phrase-beginnings in measures 5 and 7, arpeggiating through a dominant-seventh chord on IV and I, challenges this interpretation. One could decide between these possibilities by favoring one factor over another: an analyst emphasizing IOI would probably call this a 6/6 chorus; another, focusing on melodic continuity, would group the parallel beginnings in measures 5 and 7 together as part of a larger phrase, and call this a 4/8 chorus.

[3.13] For statistical purposes, I classified each of the 156 choruses as one schema or another—a problematic exercise, as Examples 11a and 11b illustrate. Thus, my findings should be taken with several grains of salt. Tables 1–3 present some of my results. In terms of overall schema frequency (table 1), no schema stands out in an obvious way, though 6/6 appears slightly more often than would be expected if the schemata were distributed at random. Tables 2 and 3 address the question of whether Parker prefers certain schemata for the first or last chorus of a solo. The tables compare Parkers actual use of each schema in initial or final position with its expected use, given the overall schema frequencies in Table 1. None of the deviations from expectations are statistically significant. Another way to put this: knowing where a chorus fits in a larger solo makes it no easier to predict its phrasing schema.

Table 1. Overall frequency of each phrasing schema in the corpus

spacer

(click to enlarge)

Table 2. Frequency of each phrasing schema as the first chorus of a solo, compared to expectations derived from Table 1

spacer

(click to enlarge)

Table 3. Frequency of each phrasing schema as the last chorus of a solo, compared to expectations derived from Table 1

spacer

(click to enlarge)

[3.14] I also considered the question of whether Parker showed a preference for consistency or variety of phrasing between successive choruses. For example, if he uses the 6/6 schema in a particular chorus, is he more likely than expected to use it in the next chorus? That would suggest Parker prefers consistency from chorus to chorus. Or is it the case that, for example, if he uses the 6/6 schema in one chorus, he is less likely than expected to use it in the next chorus, disproportionately favoring other types? That would suggest he prefers variety from chorus to chorus. In fact, having considered 117 chorus-to-chorus successions, I could identify no such preferences: knowing what type of chorus Parker just played makes it no easier to predict what type of chorus he will play next.

[3.15] The absence of clear positional preferences or syntax may lead to doubts about the significance of phrasing to Parker himself. On this matter, possible points of view range from “Parker knew exactly what he was doing” to “The recurrence of these schemata is a coincidence arising from Parker’s consideration of other matters.” Surely, sometimes chorus-level phrase structures are a byproduct of other features—especially in the case of ambiguous choruses. But it is inconceivable to me that Parker did not have at least an intuitive sense of chorus-level phrase structure: for example, an awareness that all 6/6 choruses were alike in an important way, and that they embodied a certain metrical tension compared to, say, 4/4/4 choruses.(15) I believe the schemata are an effective way of understanding recurring phrase structures—Parker’s solutions to the meter problem—and that they have some analog in Parker’s mind.(16)

Melodic Schemata

Example 12. One common melodic schema: the Descent to spacer , spacer spacer spacer spacer spacer spacer spacer

spacer

(click to enlarge and listen)

[4.1] Though Parker is free to deviate from the blues harmonic structure, he tends to work within it. This poses the harmony problem: how to improvise varied melodies roughly following the harmony, without the need to compose from scratch each time. I propose that Parker solved this problem with the melodic schemata: recurring stepwise paths, spanning around one to eight measures, which a melody seems to follow.(17) Example 12 illustrates one schema commonly found in measures 4–7 of Parker’s blues, the “Descent to spacer ,” including spacer spacer spacer spacer spacer spacer spacer in its complete form. (In this and subsequent examples, schematic notes are beamed together, and “Descent” is abbreviated to “D.”) Note that the path is interrupted several times by intervening notes.(18) Schemata can also be interrupted by rests, or presented without interruption: one note after the other. (Since interrupted schemata tend to be more interesting, I focus on them here.)

[4.2] I identified schemata by observing salient stepwise paths throughout the corpus. Given the density of notes in a typical Parker solo and the stepwise voice leading built into the blues’ harmonic progression, it is possible to find many stepwise patterns, especially when interrupted schemata open the door for unsupported, “connect-the-dots” analysis. To counter this, I required that schemata contain at least three scale degrees. Furthermore, I preferred that the schematic notes on either side of an interruption be emphasized in some way. Emphases arise from duration, register, meter, dynamics, and placement within a phrase (proximity to the beginning or end). This is not a strict requirement, but it motivated me to reject many possible analyses. In the remainder of this section, I compare the schematic approach with similar approaches. Then I present four common melodic schemata.

[4.3] The schematic approach has two significant antecedents in jazz scholarship: formulaic analysis and Schenkerian analysis. In his seminal 1974 dissertation on Parker, Thomas Owens identifies approximately one hundred melodic formulas, ranging from four-note fragments to multi-measure phrases, that recur throughout Parker’s work.(19) Though they often appear in verbatim repetition, Parker also subjects the formulas to “metric displacement, augmentation and diminution, addition and subtraction of notes, and altered phrasing and articulation” (vol. 1, ix). As the raw materials of improvisation, subject to local variation, each is thus a kind of schema.(20)

Example 13. A comparison of Owens’s formulaic approach and the schematic approach, in measures 3–5 of the blues

spacer

(click to enlarge)

[4.4] Compared to Owens’s formulas, the schemata consider more abstract melodic connections. Example 13 shows Owens’s examples of the “M.10” formula,

gipoco.com is neither affiliated with the authors of this page nor responsible for its contents. This is a safe-cache copy of the original web site.